Ventilator-Associated Pneumonia (VAP) is a common, expensive and deadly illness that threatens the health of patients receiving breathing assistance. Maintaining the patient's upper body at angles elevated between 30 and 45 above horizontal is among the most well-known and effective techniques for VAP prevention and is included in all VAP prevention guidelines. Unfortunately, studies relying on automatic, periodic head-of-bed angles recordings suggest that average angles are normally less than 30, even in units participating in process improvement projects to increase compliance with this practice. We propose to address this issue from a human factors engineering framework. Specifically we seek to identify, measure and address two broad categories of human error that may be limiting the effective application of the protocol: a) lapses or slips, which occur when the person responsible for adjusting the bed has the correct intention, but forgets or incorrectly acts on the intention, and b) mistakes, which occur when the person incorrectly believes that the bed angle is correctly set when it is not in compliance with the recommended angles. We will deploy a wireless network of battery-powered angle sensors to continuously measure head of bed angles in an intensive care unit and display these angles at the bedside and at the nurses' station. Our central hypothesis is that this feedback will increase compliance with head of bed position guidelines. We will prove this hypothesis by pursuing three specific aims. The first is to identify the root causes of non-compliant head-of-bed angles. The second is to determine whether providing immediate feedback in the patient room reduces the frequency and impact of lapses and slips. The third is to determine whether providing feedback at the nurses' station and to doctors and nurse managers on hand-held devices reduces the frequency and impact of mistakes. Not only will the methods of this research improve compliance, but they will also provide a better understanding of the general obstacles to obtaining recommended bed tilt angles, particularly with respect to the role of human error. What we learn deploying the distributed sensor network will be useful in future projects to use automated, real-time, immediate feedback to improve compliance with other patient safety-related practice.